American Journal ofPathology, Vol 136, No. 5, May 1990 Copyright ) American Association ofPathologists
Immunopathology of Adrenal and Renal Cortical Tumors Coordinated Change in Antigen Expression Is Associated with Neoplastic Conversion in the Adrenal Cortex
Richard J. Cote, Carlos Cordon-Cardo, Victor E. Reuter, and Paul Peter Rosen From the Departmenit ofPathology, Memorial SloanKetteritng Canicer Center, New York, New York
A series of adrenal cortical adenomas (ACA) and carcinomas (ACC), as well as normal adrenal cortex have been studied by a panel of 11 antibodies to characterize antigenic changes that may distinguish these morphologically similar entities. Normal adrenal cortex andACA express low-molecular weight cytokeratin intermediate filaments. However, none of the six primary or seven metastatic ACCs were found to express detectable levels of cytokeratins. In contrast, vimentin was seen in all ACCs studied and was heterogeneously expressed by ACAs. However, its expression was usually confined to stromal elements of the normal adrenal cortex. We conclude that adrenal cortical cells undergo characteristic changes in intermediate filament expression during the process of neoplastic conversion and malignant transformation. Undetectable expression of cytokeratins and strong expression of vimentin is associated with malignant adrenal cortical lesions. In addition, we examined the antigenic phenotype ofa series ofprimary renal cell carcinomas (RCC). Renal cell carcinomas express cytokeratins, whileACCs do not. The majority ofprimary RCCs express Lewis blood group isoantigens (most commonly Lewis X), while ACAs and ACCs do not. Thepanel ofantibodies described here may help to distinguish morphologically similar lesions of like histogenesis (ACAs vs. ACCs) and lesions of different histogenesis (adrenal vs. renal) on the basis of their composite antigenic phenotypes. (Am JPathol 1990, 136:1077-1084)
The morphologic distinction of adrenal cortical adenomas (ACA) from adrenal cortical carcinomas (ACC) can be difficult. A number of investigators have attempted to establish criteria to distinguish these lesions. Histologic parameters that have been found most useful are presence of necrosis, nuclear atypia, mitotic rate, and vascular/capsular invasion.' 5 Weight is generally believed to be the best predictor of behavior, although tumors weighing less than 50 g have metastasized and tumors weighing more than 100 g may not recur after surgical removal.6` Tumor aneupolidy, as determined by DNA content analysis, can further substratify tumors.9 10 However, none of the criteria predicts the ultimate behavior of all tumors. Another diagnostic problem is the distinction between ACCs and renal cell carcinoma (RCC). The histogenesis of lesions involving the superior pole of the kidney and the adrenal gland can be difficult to determine because of location and similarities in histologic appearance. The differential diagnosis of metastatic tumors with clear or oncocytic cytoplasm often includes ACCs and RCCs. We tested a panel of monoclonal antibodies on ACA, ACC, and RCC to determine their respective antigenic phenotypes and found that these lesions can be distinguished antigenically. In addition, we report changes in the antigen expression of adrenal cortical neoplasms that characterize progression from benign to malignant lesions.
Materials and Methods Cases of ACA, ACC, and RCC reviewed at Memorial Sloan-Kettering Cancer Center (MSKCC) were studied. Presented in part at the 78th Annual Meeting of The International Academy of Pathology, San Francisco, California, March 7, 1989. Accepted for publication December 20,1989. Address reprint requests to Richard J. Cote, MD, Department of Pathology, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10021.
1077
1078
Cote et al
AJP May 1990, Vol. 136, No.
5
Table 1. Immunohistochemical Reagents Pretreatment/digestion Antibody 0.1% pepsin, 30 minutes AE-1 0.1% pepsin, 30 minutes CAM 5.2 0.025% trypsin, 7 minutes Vimentin 0.05% saponin, 30 minutes Chromogranin 0.05% saponin, 30 minutes pS-100 0.0025% pronase, 4 minutes 0.05% saponin, 30 minutes Blood group antigenst
Source
Antibody dilution
Sig BD Biog Hyb Dako
1/103 1/102 1/103 1/104 1/103
Sig
1/8
All reagents are mouse monoclonal antibodies except that to pS-100, which is a polyclonal rabbit antisera, and the Ulex europaeus lectin.
t Antibodies to blood group antigens A, B and Lewis a, b, Y, and X. Ulex europaeus lectin used to identify H.
Antibody Sources: Sig: Signet Labs, Dedham, MA; BD: Becton-Dickenson, San Jose, CA; Biog: Biogenex, San Ramon, CA; Hybritech, San Diego, CA; Dako, Carpinteria, CA.
Formalin-fixed (10% buffered formalin) paraffin-embedded tissues were used. Paraffin-embedded tissue blocks were obtained in those cases processed at other hospitals. The avidin-biotin peroxidase conjugate method was used with minor modifications.1' Briefly, 5-um thick tissue sections were cut onto slides coated with 0.005% poly-llysine (Sigma Chemical Co., St. Louis, MO) and incubated at 60°C for 2 hours. After rehydration the endogenous peroxidase activity was quenched by incubation in 1.0% H202 (diluted in distilled water) for 10 minutes. The slides were washed in phosphate-buffered saline and exposed to protease digestion and/or detergent,12 as indicated in Table 1. After washing in phosphate buffered saline (PBS), sections were incubated in 5% horse serum/PBS or 5% rabbit serum (in the case of antibody to pS-1 00) for 1 hour at room temperature. This was aspirated, replaced with appropriately diluted primary antibody (Table 1), and incubated in sealed moisture chambers for 12 hours at 4°C. The sections were then washed in PBS and biotinylated horse anti-mouse immunoglobulin (Vector Labs, Burlingame, CA) (or sheep anti-rabbit immunoglobulin (Vector Labs) in the case of pS-100) was incubated for 1 hour at room temperature. After washing in PBS, the sections were incubated with avidin-biotin complex (1: 100 dilution in PBS; DAKO, Santa Barbara, CA) for 1 hour at room temperature. After a final rinse with PBS, the sections were incubated in 3,3'-diaminobenzidine (0.06% in PBS) (Sigma) with 0.003% H202 for 10 to 15 minutes. After counterstaining with modified Harris hematoxylin (Fisher, Orangeburg, NY) the sections were dehydrated and coverslipped with Permount. The primary antibodies along with their working dilutions and commercial sources are listed in Table 1. Dilutions were in 2% bovine serum albumin (BSA)/PBS. Also listed in Table 1 are the enzymatic/detergent pretreatments used. In the case of reactivity with the lectin Ulex europaeus, endogenous peroxidase in tissue sections was quenched as described. Sections were then incubated with Ulex europaeus lectin (1:500 dilution in PBS, Vector Labs) for 2.5 hours at room temperature. The slides were rinsed in PBS and sections were incubated in 5% rabbit serum for 1
hour at room temperature, aspirated, and incubated with goat anti-lectin immunoglobins (Vector Labs) for 12 hours at 4°C. The remainder of the procedure is as described, with the use of biotinylated rabbit anti-goat immunoglobulins (1:1500 in PBS/BSA; Vector Labs).
Results Clinical and Pathologic Findings The clinical/pathologic features of the ACA and ACC (primary and metastatic) are given in Tables 2, 3, and 4. None of the ACAs (Table 2) has recurred, with a follow-up time ranging from 2 to 30 months (mean, 17 months). Primary ACCs (Table 3) were considered malignant either on the basis of metastases (at the time of or after diagnosis) (cases 7 to 10), direct invasion of contiguous structures (case 12), or based on size and histologic findings, particularly necrosis (case 1 1). The metastatic ACCs (Table 4) had primary tumors of the adrenal cortex documented at MSKCC or outside institutions, and in the latter cases microscopic material was reviewed at MSKCC. All cases classified as renal cell carcinoma were primary tumors and all had clear or tubular cell features.
Immunohistochemical Results (Table 5) Cytokeratins Two widely available antibodies that recognize lowmolecular weight cytokeratins were used (AE-1, Cam Table 2. Adrenal CorticalAdenomas Tumor
Case 1
2 3 4 5 6
Age 79 39 71
40 57 50
Sex F F F M
M M
weight (g) 60 18 19 66 24 73
Clinical presentation
Sarcoma* t 17 ketosteroids NS Mass Mass Mass
NS, not stated. * Preoperative diagnosis based on fine-needle aspiration.
Adrenal and Renal Cortical Tumors
1079
AJP May 1990, Vol. 136, No. 5
Table 3. Primary Adrenal Cortical Carcinomas
Case 7
Age
Sex
Tumor weight (g)
Clinical
63
M
2400
t 1 7-ketosteroids
8 9 10
41 17 49
F
M M
960 410 1100
NS Cushings NS
11
63
M
450
t 17 ketosteroids
12
30
F
900
NS
Comments Mets present at primary resection Mets at 2 years Liver and lung mets Mets present at primary resection Necrosis, complete replacement adrenal gland Invasion of liver
NS, not stated; Mets, metastasis.
5.2). Cytokeratin expression was seen in all examples (four of four) of normal adrenal cortex (Figure 1A). The staining was observed to be focal and generally restricted to the zona glomerulosa. All examples (six of six) of ACAs showed expression of cytokeratins as well. However, a range of expression was noted. While some tumors demonstrated a strong staining pattern (Figure 1 B), others showed weak staining, with only rare cells positive (Figure 1 C). While all ACAs studied had focal reactivity with Cam 5.2, only two of six cases exhibited detectable expression of the epitopes recognized by antibody AE-1. In contrast, none of the 13 ACCs had detectable staining with either AE-1 or Cam 5.2 (Figure 1 D). All examples of the primary renal cortical carcinomas (17 of 17) demonstrated expression of cytokeratins, in most cases staining with both AE1 (12 of 17) and Cam 5.2 (17 of 17) (Figure 1E). Protease digestion was used in all cases to enhance immunoreactivity, as outlined in Materials and Methods.
Vimentin Vimentin expression in the normal adrenal cortex was primarily confined to stromal elements, with only questionable reactivity seen in adrenal cortical cells (Figure 2A). Adrenal cortical adenomas exhibited a range of expression, with some tumors showing staining mainly in stromal elements (Figure 2B), while others had strong expression Table 4. Metastatic Adrenal Cortical Carcinomas No. of metastatic sites Case Sex studied Age M 1 7* 63 8* 44 F 1 M 10* 49 2t M 13 63 4t 14 F 51 2t M 15 45 1 16 52 F 2t Primary site immunophenotyped.
t All metastatic sites were immunophenotypically similar.
of vimentin in the neoplastic cells (Figure 2C). Vimentin expression was observed in all (13 of 13) ACCs studied, including all primary and metastatic lesions (Figure 2D). The staining was generally strong and fairly uniform. In the case of renal cell carcinomas, 5 of 17 primary lesions exhibited focal expression of vimentin.
pS-100 Expression of S-100 protein was observed in normal adrenal cortex (3 of 4), ACAs (2 of 3), and ACCs (8 of 12), as well as in the majority of renal cell carcinomas (16 of 1 7).
Chromogranin Chromogranin expression was not observed in any of the adrenal cortical lesions or examples of normal adrenal cortex studied. However, the adrenal medulla showed strong expression of choromogranin in the normal adrenal glands studied.
Blood Group Antigens Monoclonal antibodies directed against the major blood group antigens A and B, as well as those against Lewis a, b, X, and Y were studied. In addition, the lectin Ulex europaeus was used to define blood group H expression. Cells derived from adrenal cortex (normal adrenal cortex, ACAs, and ACCs) were found to lack detectable expression of all blood group antigens studied (Figures 3A and B). None of the RCCs studied showed detectable expression of the major blood group antigens (A, B, and H), but 13 of 17 RCCs exhibited expression of Lewis-type antigens: Lewis X (12 of 17), Lewis a (3 of 17), and Lewis b (1 of 17) (Figure 3C). None showed expression of Lewis Y.
Expression of Cytokeratin and Vimentin in ACA As previously noted, ACAs showed a range of staining for both cytokeratins and vimentin. Table 6 shows that the
1080
Cote et al
AJP May 1990, Vol. 136, No. 5
Table 5. Antigen Expression in Adrenal Cortical Neoplasms and Renal Cell Carcinomas Chromogranin Vimentin Cam 5.2 AE-1 0/4 0/4t 1/4 Normal adrenal cortex 4/4t 0/6 4/6t 6/6t 2/6 ACA 0/6 6/6 0/6 0/6 ACC primary 0/7 7/7 0/7 0/7 metastases ND 5/17 17/17 12/17 RCC
S-100
BGA*
3/4 2/3 5/6 3/6 16/17
0/4 0/3 0/6 0/6 13/17§
BGA: Blood group antigens A, B, H, and Lewis a, b, X, and Y.
t Focal expression in all cases. t Stromal elements stained in all cases. § Lewis a, b or X positive; A, B, H, and Lewis Y negative. ND, not done.
expression of these two classes of intermediate filament proteins appears to be inversely related in individual lesions. In particular, cases 3 and 5 had strong focal expression of cytokeratins (Figure 1B), while the expression of vimentin in each of these cases was confined primarily to stromal elements (Figure 2B). Conversely, case 1 showed weak expression of cytokeratins, with only a few cells reactive, while the expression of vimentin in this case was strong and fairly uniform (Figures 1 C and 2C). We did not observe any consistent morphologic differences between these antigenically distinct ACAs. However, the tumor that exhibited the weakest expression for cytokeratin (and strong expression for vimentin) was one of the larger ACAs (case 1, 60 g).
Discussion When faced with a moderately large but apparently wellcircumscribed adrenal cortical tumor, the distinction between adenoma and carcinoma can be difficult. A series of well-defined histologic characteristics, including vascular/capsular invasion, necrosis, nuclear pleomorphism, and architectural features have been developed to aid in this distinction.1-5 More recently the DNA content of the neoplastic cells comprising the tumor has been shown to separate lesions with more aggressive behavior.9 10 However, none of these criteria can predict outcome in all cases. In this study we have used a panel of commonly available monoclonal antibodies to define the antigenic phenotype of ACAs and ACCs. Previous reports have indicated that no clear distinctions could be made based on the antibodies used.13 14 Miettinen et al13 found that tumors arising in the adrenal cortex could be distinguished from those arising in the medulla on the basis of intermediate filament expression, but observed no consistent distinctions between ACAs and ACCs. Wick et al,14 using a more extended panel of antibodies, also concluded that there were no clear differences in the antigenic phenotypes of ACAs and
ACCs.
In contrast to the results obtained in the current study, these investigators found that a proportion of ACCs express cytokeratins.13,14 In the study of Miettinen et al,13 both frozen and paraffin-embedded tissues were used and there were no obvious differences between these types of tissue preservation. Three different monoclonal antibodies to cytokeratins, developed by that group, were used. Wick et al14 used paraffin-embedded tissues in their study and found that, even with protease treatment, few ACCs expressed detectable levels of cytokeratins. A common feature of both these studies was that relatively few ACCs expressed cytokeratins, usually at decreased levels. In addition, Miettinen et al13 reported that increased expression of vimentin was seen in ACC, compared with ACA or normal adrenal cortex, which is consistent with our findings. We have found that in the cases studied here there is consistent and distinctive pattern of antigen expression in comparing normal adrenal cortex, ACAs, and ACCs (Table 5). All examples of normal cortex studied showed cytokeratin expression, but only minimal (or equivocal) expression of vimentin, generally confined to stromal elements (Figures 1 A and 2A). The majority of ACAs studied expressed cytokeratin; a range of expression was noted, with some tumors showing strong staining and others weak and focal staining (Figures 1 B and C). Adrenal cortical adenomas also showed a range of expression of vimentin, with some tumors having weak/minimal expression, while others exhibit strong reactivity (Figures 2B and C). Adrenal cortical adenomas tend to show expression of both cytokeratins and vimentin (Tables 5 and 6). None of the 13 ACCs studied, including examples of primary and metastatic lesions, showed detectable expression of cytokeratins. However, all showed strong, fairly uniform reactivity for vimentin (Figures 1 D and 2D). There appears to be a coordinated change in intermediate filament expression in comparing normal adrenal cortex, ACA, and ACC. Normal cortex expresses cytokeratins, primarily in the zona glomerulosa, but only minimally expresses vimentin. Adrenal cortical carcinomas show the reverse; none showed reactivity for cytokeratins, but
Adrenal and Renal Cortical Tumors
1081
AJP May 1990, Vol. 136, No. 5
;l. a. r:ll
)
'@
*
; } :t
.
s
@
.j
.e
fi
@
S11
2
f;
,',."
Xlv j
jT
*
i
< N ;W:
UP
Figure 1. Reactivity with antibody to cytokeratins (Cam 5.2) (X250). A: Normal adrenal cortex, positive reactivity; B: Adrenal cortical adenoma (case 3). Note strongfocal reactivity; C: Adrenal cortical adenoma (case 1). Note weak staining offew cells. D: Adrenal cortical carcinoma (case 12). No reactivity seen. E: Renal cell carcinoma, positive reactivity. Figure 2. Reactivity with antibody to vimentin (X250). A: Normal adrenal cortex, staining confined to stromal elements; B: Adrenal cortical adenoma (case 3). Stainingprimarily confined to stromal elements: C: Adrenal cortical adenoma (case 1). Note strong reactivity; D: Adrenal cortical carcinoma (case 12), positive reactivity.
1082
Cote et al
AJPMay 1990, Vol. 136, No. 5
..
P
.
t4i; !
11~~~
-1
,
.! OF
-
,
k..~~U
&
-i-.t1f4
Figure 3.
Reactiv)ity
with antibodies to blood group
adrenal cortical carcinoma,
no
antigens (Lewis X) (X250). A: Adrenal cortical adenoma, reactivity seen; C: renal cell carcinoma, positive reactivity.
all expressed vimentin. There is a range of expression of both cytokeratins and vimentin in ACA, a characteristic not seen in normal cortex or ACC. In ACA, this range of intermediate filament expression appears to be coordinated in individual lesions. Adrenal cortical adenomas that show strong reactivity for cytokeratins only weakly express vimentin, while those that are weakly reactive for cytokeratins show strong expression of vimentin (Table 6 and Figures 1 B, 2B and 1C, 2C). These results suggest that adrenal cortical cells undergo characteristic changes in intermediate filament expression during the process of neoplastic conversion and malignant transformation. Our data would indicate that neoplastic conversion is associated with increased vimentin expression and decreased cytokeratin expression at the cellular level. Furthermore, it appears that using the techniques described here, lack of cytokeratin expression and strong vimentin expression is associated with malignant lesions. It is well known that malignant tumors of epithelial origin can coexpress both cytokeratins and vimentin.152 Furthermore, increased vimentin expression is associated Table 6. Comparison of Cytokeratin and Vimentin Expression in Adrenal Cortical Adenomas Cam 5.2 Case Vimentin* 1 + few cells +++ many cells 2 ++ focal +-++ focal +++ focal - most cells (+ stroma) 3 4 ++ focal + focal ++-+++ focal - most cells (+ stroma) 5 + focal ++ focal 6 Staining intensity: +, weak; ++, moderate; +++, strong; * stromal/ vascular elements stained in all cases.
no
reactivity
seen; B:
with more highly anaplastic epithelial tumors and anaplastic areas within these tumors, and thus it is correlated with certain morphologic features related to more aggressive behavior."5' However, we are not aware of any previous study or system in which the coordinated changes in intermediate filament expression described here were observed throughout the range of neoplastic growth. In addition, conversion from cytokeratin to vimentin expression is normally associated with morphologic changes in most tumors, primarily of the metaplastic or spindle cell type.15- " This was not the case for adrenal cortical neoplasms, where increased vimentin and decreased cytokeratin expression was not associated with any characteristic morphologic changes. The antigenic phenotype of adrenal cortical hyperplasia appears to be distinct and unrelated to this neoplastic conversion because none of four hyperplastic nodules studied had detectable expression of either cytokeratin or vimentin (unpublished data). As previously noted, cytokeratin expression is principally seen in the zona glomerulosa, while vimentin is not seen well in any layers of the adrenal cortex. This may indicate that adrenal cortical hyperplasia (those studied here at any rate) may arise from cells distinct from those that give rise to adrenal cortical neoplasms, and that hyperplasia may not be,part of the pathway of transformation in the adrenal cortex. The second major issue investigated here was whether antibody panels could distinguish adrenal cortical lesions from RCC. These tumors can be quite similar morphologically, and they are included in the differential diagnosis of metastatic lesions that have clear or granular cytoplasm. In addition, the primary site of a tumor involving both the adrenal gland and upper pole of the kidney
Adrenal and Renal Cortical Tumors
1083
AJP May 1990, Vol. 136, No. 5
can be difficult to determine, even after detailed microscopic examination. Our results suggest that the most useful antibodies are those with specificity to cytokeratins and blood group antigens (Table 5). Adrenal cortical carcinomas did not express detectable levels of cytokeratins. Cytokeratins were demonstrated in all primary clear cell and tubular RCCs examined. The majority of primary RCCs express blood group antigens, but all ACAs and ACCs were negative (Table 5). Reactivity with antibodies to vimentin and protein S-100 do not contribute to the differential diagnosis. Although Wick et al14 found that ACCs and RCCs did not express detectable levels of protein S-100, we found that many ACCs and the majority of RCCs showed positive reactivity to the antibody to protein S-100. Expression of protein S-100 by normal and malignant renal epithelium has been previously reported.21 22 Wick et al14 also demonstrated that the majority of RCCs expressed cytokeratins as well. In addition, they found that RCCs expressed epithelial membrane antigen (EMA), while ACAs and ACCs were negative. They reported that RCCs reacted with antibodies to the major blood group isoantigens (A, B, and H),14 but in our series none of the RCCs tested expressed detectable levels of these antigens. However, the majority of primary RCCs expressed Lewis-type blood group antigens, most commonly Lewis X, and occasionally Lewis a or b.23 One possible reason for this discrepancy is that the antibodies to the major blood group antigens used in the previous study may have had unsuspected cross reactivity with Lewistype antigens as well. It is important to emphasize the fact that the antigenic phenotype for RCCs reported here pertains to primary renal cortical tumors with clear cell or granular cell features. We have found that metastatic RCC less commonly express Lewis blood group antigens.23 In addition, some metastatic RCCs express low to undetectable levels of cytokeratins and high levels of vimentin (Cordon-Cardo et al, unpublished data), thus making their antigenic phenotype more similar to that of ACC. However, RCCs that lack expression of cytokeratins are generally poorly differentiated tumors, often with pronounced spindle cell features,'4 and thus can generally be separated from ACCs on morphologic grounds. Finally, RCCs and adrenal cortical lesions can be distinguished by a series of antigens that are expressed by cells of proximal tubular origin.24' These antigens are, however, only stable in frozen sections and were not used in the current study. This study points out once again the importance of using antibody panels in evaluating lesions of uncertain histogenesis or malignant potential. While none of the antigens studied here are specific for adrenal cortical or renal cells, lesions of similar histogenesis (ACAs vs. ACCs)
and lesions of different histogenesis (adrenal cortex vs. renal cortex) can be distinguished on the basis of their composite antigenic phenotypes. An important future step will be to determine if those features that appear to distinguish ACAs from ACCs (primarily cytokeratin and vimentin expression) correlate with more established criteria (morphology, weight, and DNA content analysis), particularly for borderline lesions. We expect that, given more sensitive detection systems and a greater number of tumors studied, cytokeratin expression will be shown in at least a proportion of ACCs. However, the finding of strong vimentin expression associated with weak cytokeratin expression should raise the possibility of malignant potential in an adrenal cortical tumor.
References 1. Heinbecker P, O'Neal LA, Ackerman LV: Functioning and nonfunctioning adrenal cortical tumors. Surg Gynecol Obstet 1957,105:21-33 2. Huvos AG, Hajdu SI, Brasfield RD, Foote FW: Adrenal cortical carcinoma: Clinicopathologic study of 34 cases. Cancer
1969.25:354-361 3. Kay S: Hyperplasia and neoplasia of the adrenal gland. Pathol Annu 1976,11:103-139 4. King DR, Lack EE: Adrenal cortical carcinoma: A clinical and pathologic study of 49 cases. Cancer 1979,44:239-244 5. Weiss LM: Comparative histologic study of 43 metastasizing and nonmetastasizing adrenal cortical tumors. Am J Surg Pathol 1984, 8:163-169 6. Schteingart DE, Oberman HA, Friedman BA, Conn JW: Adrenal cortical neoplasms producing Cushings syndrome. Cancer 1968, 22:1005-1013 7. Tang CK, Gray GF: Adrenal cortical neoplasms: Prognosis and morphology. Urology 1975, 5:691-695 8. Hough AJ, Holifield JW, Page DL, Hartmann WH: Prognostic factors in adrenal cortical tumors: A mathematical analysis of clinical and morphologic data. Am J Clin Pathol 1979, 72: 390-399 9. Amberson JB, Vaughn ED, Gray GF, Naus GJ: Flow cytometric analysis of nuclear DNA from adrenal cortical neoplasms: A retrospective study using paraffin-embedded tissue. Cancer 1987, 59:2091-2095 10. Taylor SR, Roederer M, Murphy RF: Flow cytometric DNA analysis of adrenocortical tumors in children. Cancer 1987, 57:2059-2063 11. Hsu S-M, Raine L, Fanger H: Use of avidin-biotin-peroxidase complex (ABC) in immunoperoxidase techniques: A comparison between ABC and unlabelled antibody (PAP) procedures. J Histochem Cytochem 1981,29:577-580 12. Cordon-Cardo C, Finstad CL, Bander NJ, Melamed MR: Immunoanatomic distribution of cytostructural and tissue-associated antigens in the human urinary tract. Am J Pathol
1987,126:269-284 13. Miettinen M, Lehto VP, Virtanen I: Immunofluorescence microscopic evaluation of the intermediate filament expression
1084
Cote et al
AJP May 1990, Vol. 136, No. 5
14.
15.
16.
17.
18.
19.
20.
21.
of the adrenal cortex and medulla and their tumors. Am J Pathol 1985,118:360-366 Wick MR, Cherwitz DL, McGlennen RC, Dehner LP: Adrenocortical carcinoma: An immunohistochemical comparison with renal cell carcinoma. Am J Pathol 1986; 122:343-352 Medeiros LJ, Michie SA, Johnson DA, Warnke RA, Weiss LM: An immunoperoxidase study of renal cell carcinomas: Correlation with nuclear grade, cell type, and histologic pattern. Human Pathol 1988,19:980-987 Oberman HA: Metaplastic carcinoma of the breast: A clinicopathologic study of 29 patients. Am J Surg Pathol 1987,11: 918-929 Meis JM, Ord6onez NG, Gallager HS: Sarcomatoid carcinoma of the breast: An immunohistochemical study of six cases. Virch Arch 1987, 410:415-421 Santensanio G, Pascal RR, Bisceglia M, Costantino AM, Bosman C: Metaplastic breast carcinoma with epithelial phenotype and pseudosarcomatous components. Arch Pathol Lab Med 1988,112:82-85 Hurlimanh J, Gardiol D, Scazziga B: Immunohistology of anaplastic thyroid carcinoma: A study of 43 cases. Histopathology 1987,11:567-580 LiVolsi VA, Brooks JJ, Arendash-Durand B: Anaplastic thyroid tumors: Immunohistology. Am J Clin Pathol 1987, 87: 434-442 Vanstapel M-J, Gatter KC, Woolf-Peeters C, Mason DY, Desmet VD: New sites of human S-100 immunoreactivity detected with monoclonal antibodies. Am J Clin Pathol 1986,
85:160-168
22. Kato K, Haimoto H, Ariyoshi Y, Horisawa M, Washida H, Kimura S: High levels of S-100 protein in tumor tissues and in serum of patients with renal cell carcinoma. Jpn J Cancer Res 1985, 76:856-862 23. Cordon-Cardo C, Reuter VE, Finstad CL, Sheinfeld J, Lloyd KO, Fair WR, Melamed MR: Blood group-related antigens in human kidney: Modulation of Lewis determinants in renal cell carcinoma. Cancer Res 1989,49:212-218 24. Cordon-Cardo C, Bander NH, Fradet Y, Finstad CL, Whitmore WF, Lloyd KO, Oettgen HF, Melamed MR, Old LJ: Immunoanatomic dissection of the human urinary tract by monoclonal antibodies. J Histochem Cytochem 1984, 32:
1035-1040 25. Bander NH, Cordon-Cardo C. Monoclonal antibodies to renal cell markers. In Sell S, Reisfeld RA, eds. Monoclonal Antibodies in Cancer. Clifton, NJ, Humana Press, 1985. pp. 325-338. 26. Finstad CL, Cordon-Cardo C, Bander NH, Whitmore WF, Melamed MR, Old LJ: Specificity analysis of mouse monoclonal antibodies defining cell surface antigens of human renal cancer. Proc Natl Acad Scie USA 1985, 82:2955-2959
Acknowledgments The authors thank Drs. Myron Melamed and Murray Brennan for their comments. Daisy Jimenez-Joseph and Mary Ann Gangi provided expert technical assistance, Kin Kong and Lisa Hollis prepared the photomicrographs, and Grace Bell and Alice Haynes prepared the manuscript.
Immunopathology of Adrenal and Renal Cortical Tumors Coordinated Change in Antigen Expression Is Associated with Neoplastic Conversion in the Adrenal Cortex
Richard J. Cote, Carlos Cordon-Cardo, Victor E. Reuter, and Paul Peter Rosen From the Departmenit ofPathology, Memorial SloanKetteritng Canicer Center, New York, New York
A series of adrenal cortical adenomas (ACA) and carcinomas (ACC), as well as normal adrenal cortex have been studied by a panel of 11 antibodies to characterize antigenic changes that may distinguish these morphologically similar entities. Normal adrenal cortex andACA express low-molecular weight cytokeratin intermediate filaments. However, none of the six primary or seven metastatic ACCs were found to express detectable levels of cytokeratins. In contrast, vimentin was seen in all ACCs studied and was heterogeneously expressed by ACAs. However, its expression was usually confined to stromal elements of the normal adrenal cortex. We conclude that adrenal cortical cells undergo characteristic changes in intermediate filament expression during the process of neoplastic conversion and malignant transformation. Undetectable expression of cytokeratins and strong expression of vimentin is associated with malignant adrenal cortical lesions. In addition, we examined the antigenic phenotype ofa series ofprimary renal cell carcinomas (RCC). Renal cell carcinomas express cytokeratins, whileACCs do not. The majority ofprimary RCCs express Lewis blood group isoantigens (most commonly Lewis X), while ACAs and ACCs do not. Thepanel ofantibodies described here may help to distinguish morphologically similar lesions of like histogenesis (ACAs vs. ACCs) and lesions of different histogenesis (adrenal vs. renal) on the basis of their composite antigenic phenotypes. (Am JPathol 1990, 136:1077-1084)
The morphologic distinction of adrenal cortical adenomas (ACA) from adrenal cortical carcinomas (ACC) can be difficult. A number of investigators have attempted to establish criteria to distinguish these lesions. Histologic parameters that have been found most useful are presence of necrosis, nuclear atypia, mitotic rate, and vascular/capsular invasion.' 5 Weight is generally believed to be the best predictor of behavior, although tumors weighing less than 50 g have metastasized and tumors weighing more than 100 g may not recur after surgical removal.6` Tumor aneupolidy, as determined by DNA content analysis, can further substratify tumors.9 10 However, none of the criteria predicts the ultimate behavior of all tumors. Another diagnostic problem is the distinction between ACCs and renal cell carcinoma (RCC). The histogenesis of lesions involving the superior pole of the kidney and the adrenal gland can be difficult to determine because of location and similarities in histologic appearance. The differential diagnosis of metastatic tumors with clear or oncocytic cytoplasm often includes ACCs and RCCs. We tested a panel of monoclonal antibodies on ACA, ACC, and RCC to determine their respective antigenic phenotypes and found that these lesions can be distinguished antigenically. In addition, we report changes in the antigen expression of adrenal cortical neoplasms that characterize progression from benign to malignant lesions.
Materials and Methods Cases of ACA, ACC, and RCC reviewed at Memorial Sloan-Kettering Cancer Center (MSKCC) were studied. Presented in part at the 78th Annual Meeting of The International Academy of Pathology, San Francisco, California, March 7, 1989. Accepted for publication December 20,1989. Address reprint requests to Richard J. Cote, MD, Department of Pathology, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10021.
1077
1078
Cote et al
AJP May 1990, Vol. 136, No.
5
Table 1. Immunohistochemical Reagents Pretreatment/digestion Antibody 0.1% pepsin, 30 minutes AE-1 0.1% pepsin, 30 minutes CAM 5.2 0.025% trypsin, 7 minutes Vimentin 0.05% saponin, 30 minutes Chromogranin 0.05% saponin, 30 minutes pS-100 0.0025% pronase, 4 minutes 0.05% saponin, 30 minutes Blood group antigenst
Source
Antibody dilution
Sig BD Biog Hyb Dako
1/103 1/102 1/103 1/104 1/103
Sig
1/8
All reagents are mouse monoclonal antibodies except that to pS-100, which is a polyclonal rabbit antisera, and the Ulex europaeus lectin.
t Antibodies to blood group antigens A, B and Lewis a, b, Y, and X. Ulex europaeus lectin used to identify H.
Antibody Sources: Sig: Signet Labs, Dedham, MA; BD: Becton-Dickenson, San Jose, CA; Biog: Biogenex, San Ramon, CA; Hybritech, San Diego, CA; Dako, Carpinteria, CA.
Formalin-fixed (10% buffered formalin) paraffin-embedded tissues were used. Paraffin-embedded tissue blocks were obtained in those cases processed at other hospitals. The avidin-biotin peroxidase conjugate method was used with minor modifications.1' Briefly, 5-um thick tissue sections were cut onto slides coated with 0.005% poly-llysine (Sigma Chemical Co., St. Louis, MO) and incubated at 60°C for 2 hours. After rehydration the endogenous peroxidase activity was quenched by incubation in 1.0% H202 (diluted in distilled water) for 10 minutes. The slides were washed in phosphate-buffered saline and exposed to protease digestion and/or detergent,12 as indicated in Table 1. After washing in phosphate buffered saline (PBS), sections were incubated in 5% horse serum/PBS or 5% rabbit serum (in the case of antibody to pS-1 00) for 1 hour at room temperature. This was aspirated, replaced with appropriately diluted primary antibody (Table 1), and incubated in sealed moisture chambers for 12 hours at 4°C. The sections were then washed in PBS and biotinylated horse anti-mouse immunoglobulin (Vector Labs, Burlingame, CA) (or sheep anti-rabbit immunoglobulin (Vector Labs) in the case of pS-100) was incubated for 1 hour at room temperature. After washing in PBS, the sections were incubated with avidin-biotin complex (1: 100 dilution in PBS; DAKO, Santa Barbara, CA) for 1 hour at room temperature. After a final rinse with PBS, the sections were incubated in 3,3'-diaminobenzidine (0.06% in PBS) (Sigma) with 0.003% H202 for 10 to 15 minutes. After counterstaining with modified Harris hematoxylin (Fisher, Orangeburg, NY) the sections were dehydrated and coverslipped with Permount. The primary antibodies along with their working dilutions and commercial sources are listed in Table 1. Dilutions were in 2% bovine serum albumin (BSA)/PBS. Also listed in Table 1 are the enzymatic/detergent pretreatments used. In the case of reactivity with the lectin Ulex europaeus, endogenous peroxidase in tissue sections was quenched as described. Sections were then incubated with Ulex europaeus lectin (1:500 dilution in PBS, Vector Labs) for 2.5 hours at room temperature. The slides were rinsed in PBS and sections were incubated in 5% rabbit serum for 1
hour at room temperature, aspirated, and incubated with goat anti-lectin immunoglobins (Vector Labs) for 12 hours at 4°C. The remainder of the procedure is as described, with the use of biotinylated rabbit anti-goat immunoglobulins (1:1500 in PBS/BSA; Vector Labs).
Results Clinical and Pathologic Findings The clinical/pathologic features of the ACA and ACC (primary and metastatic) are given in Tables 2, 3, and 4. None of the ACAs (Table 2) has recurred, with a follow-up time ranging from 2 to 30 months (mean, 17 months). Primary ACCs (Table 3) were considered malignant either on the basis of metastases (at the time of or after diagnosis) (cases 7 to 10), direct invasion of contiguous structures (case 12), or based on size and histologic findings, particularly necrosis (case 1 1). The metastatic ACCs (Table 4) had primary tumors of the adrenal cortex documented at MSKCC or outside institutions, and in the latter cases microscopic material was reviewed at MSKCC. All cases classified as renal cell carcinoma were primary tumors and all had clear or tubular cell features.
Immunohistochemical Results (Table 5) Cytokeratins Two widely available antibodies that recognize lowmolecular weight cytokeratins were used (AE-1, Cam Table 2. Adrenal CorticalAdenomas Tumor
Case 1
2 3 4 5 6
Age 79 39 71
40 57 50
Sex F F F M
M M
weight (g) 60 18 19 66 24 73
Clinical presentation
Sarcoma* t 17 ketosteroids NS Mass Mass Mass
NS, not stated. * Preoperative diagnosis based on fine-needle aspiration.
Adrenal and Renal Cortical Tumors
1079
AJP May 1990, Vol. 136, No. 5
Table 3. Primary Adrenal Cortical Carcinomas
Case 7
Age
Sex
Tumor weight (g)
Clinical
63
M
2400
t 1 7-ketosteroids
8 9 10
41 17 49
F
M M
960 410 1100
NS Cushings NS
11
63
M
450
t 17 ketosteroids
12
30
F
900
NS
Comments Mets present at primary resection Mets at 2 years Liver and lung mets Mets present at primary resection Necrosis, complete replacement adrenal gland Invasion of liver
NS, not stated; Mets, metastasis.
5.2). Cytokeratin expression was seen in all examples (four of four) of normal adrenal cortex (Figure 1A). The staining was observed to be focal and generally restricted to the zona glomerulosa. All examples (six of six) of ACAs showed expression of cytokeratins as well. However, a range of expression was noted. While some tumors demonstrated a strong staining pattern (Figure 1 B), others showed weak staining, with only rare cells positive (Figure 1 C). While all ACAs studied had focal reactivity with Cam 5.2, only two of six cases exhibited detectable expression of the epitopes recognized by antibody AE-1. In contrast, none of the 13 ACCs had detectable staining with either AE-1 or Cam 5.2 (Figure 1 D). All examples of the primary renal cortical carcinomas (17 of 17) demonstrated expression of cytokeratins, in most cases staining with both AE1 (12 of 17) and Cam 5.2 (17 of 17) (Figure 1E). Protease digestion was used in all cases to enhance immunoreactivity, as outlined in Materials and Methods.
Vimentin Vimentin expression in the normal adrenal cortex was primarily confined to stromal elements, with only questionable reactivity seen in adrenal cortical cells (Figure 2A). Adrenal cortical adenomas exhibited a range of expression, with some tumors showing staining mainly in stromal elements (Figure 2B), while others had strong expression Table 4. Metastatic Adrenal Cortical Carcinomas No. of metastatic sites Case Sex studied Age M 1 7* 63 8* 44 F 1 M 10* 49 2t M 13 63 4t 14 F 51 2t M 15 45 1 16 52 F 2t Primary site immunophenotyped.
t All metastatic sites were immunophenotypically similar.
of vimentin in the neoplastic cells (Figure 2C). Vimentin expression was observed in all (13 of 13) ACCs studied, including all primary and metastatic lesions (Figure 2D). The staining was generally strong and fairly uniform. In the case of renal cell carcinomas, 5 of 17 primary lesions exhibited focal expression of vimentin.
pS-100 Expression of S-100 protein was observed in normal adrenal cortex (3 of 4), ACAs (2 of 3), and ACCs (8 of 12), as well as in the majority of renal cell carcinomas (16 of 1 7).
Chromogranin Chromogranin expression was not observed in any of the adrenal cortical lesions or examples of normal adrenal cortex studied. However, the adrenal medulla showed strong expression of choromogranin in the normal adrenal glands studied.
Blood Group Antigens Monoclonal antibodies directed against the major blood group antigens A and B, as well as those against Lewis a, b, X, and Y were studied. In addition, the lectin Ulex europaeus was used to define blood group H expression. Cells derived from adrenal cortex (normal adrenal cortex, ACAs, and ACCs) were found to lack detectable expression of all blood group antigens studied (Figures 3A and B). None of the RCCs studied showed detectable expression of the major blood group antigens (A, B, and H), but 13 of 17 RCCs exhibited expression of Lewis-type antigens: Lewis X (12 of 17), Lewis a (3 of 17), and Lewis b (1 of 17) (Figure 3C). None showed expression of Lewis Y.
Expression of Cytokeratin and Vimentin in ACA As previously noted, ACAs showed a range of staining for both cytokeratins and vimentin. Table 6 shows that the
1080
Cote et al
AJP May 1990, Vol. 136, No. 5
Table 5. Antigen Expression in Adrenal Cortical Neoplasms and Renal Cell Carcinomas Chromogranin Vimentin Cam 5.2 AE-1 0/4 0/4t 1/4 Normal adrenal cortex 4/4t 0/6 4/6t 6/6t 2/6 ACA 0/6 6/6 0/6 0/6 ACC primary 0/7 7/7 0/7 0/7 metastases ND 5/17 17/17 12/17 RCC
S-100
BGA*
3/4 2/3 5/6 3/6 16/17
0/4 0/3 0/6 0/6 13/17§
BGA: Blood group antigens A, B, H, and Lewis a, b, X, and Y.
t Focal expression in all cases. t Stromal elements stained in all cases. § Lewis a, b or X positive; A, B, H, and Lewis Y negative. ND, not done.
expression of these two classes of intermediate filament proteins appears to be inversely related in individual lesions. In particular, cases 3 and 5 had strong focal expression of cytokeratins (Figure 1B), while the expression of vimentin in each of these cases was confined primarily to stromal elements (Figure 2B). Conversely, case 1 showed weak expression of cytokeratins, with only a few cells reactive, while the expression of vimentin in this case was strong and fairly uniform (Figures 1 C and 2C). We did not observe any consistent morphologic differences between these antigenically distinct ACAs. However, the tumor that exhibited the weakest expression for cytokeratin (and strong expression for vimentin) was one of the larger ACAs (case 1, 60 g).
Discussion When faced with a moderately large but apparently wellcircumscribed adrenal cortical tumor, the distinction between adenoma and carcinoma can be difficult. A series of well-defined histologic characteristics, including vascular/capsular invasion, necrosis, nuclear pleomorphism, and architectural features have been developed to aid in this distinction.1-5 More recently the DNA content of the neoplastic cells comprising the tumor has been shown to separate lesions with more aggressive behavior.9 10 However, none of these criteria can predict outcome in all cases. In this study we have used a panel of commonly available monoclonal antibodies to define the antigenic phenotype of ACAs and ACCs. Previous reports have indicated that no clear distinctions could be made based on the antibodies used.13 14 Miettinen et al13 found that tumors arising in the adrenal cortex could be distinguished from those arising in the medulla on the basis of intermediate filament expression, but observed no consistent distinctions between ACAs and ACCs. Wick et al,14 using a more extended panel of antibodies, also concluded that there were no clear differences in the antigenic phenotypes of ACAs and
ACCs.
In contrast to the results obtained in the current study, these investigators found that a proportion of ACCs express cytokeratins.13,14 In the study of Miettinen et al,13 both frozen and paraffin-embedded tissues were used and there were no obvious differences between these types of tissue preservation. Three different monoclonal antibodies to cytokeratins, developed by that group, were used. Wick et al14 used paraffin-embedded tissues in their study and found that, even with protease treatment, few ACCs expressed detectable levels of cytokeratins. A common feature of both these studies was that relatively few ACCs expressed cytokeratins, usually at decreased levels. In addition, Miettinen et al13 reported that increased expression of vimentin was seen in ACC, compared with ACA or normal adrenal cortex, which is consistent with our findings. We have found that in the cases studied here there is consistent and distinctive pattern of antigen expression in comparing normal adrenal cortex, ACAs, and ACCs (Table 5). All examples of normal cortex studied showed cytokeratin expression, but only minimal (or equivocal) expression of vimentin, generally confined to stromal elements (Figures 1 A and 2A). The majority of ACAs studied expressed cytokeratin; a range of expression was noted, with some tumors showing strong staining and others weak and focal staining (Figures 1 B and C). Adrenal cortical adenomas also showed a range of expression of vimentin, with some tumors having weak/minimal expression, while others exhibit strong reactivity (Figures 2B and C). Adrenal cortical adenomas tend to show expression of both cytokeratins and vimentin (Tables 5 and 6). None of the 13 ACCs studied, including examples of primary and metastatic lesions, showed detectable expression of cytokeratins. However, all showed strong, fairly uniform reactivity for vimentin (Figures 1 D and 2D). There appears to be a coordinated change in intermediate filament expression in comparing normal adrenal cortex, ACA, and ACC. Normal cortex expresses cytokeratins, primarily in the zona glomerulosa, but only minimally expresses vimentin. Adrenal cortical carcinomas show the reverse; none showed reactivity for cytokeratins, but
Adrenal and Renal Cortical Tumors
1081
AJP May 1990, Vol. 136, No. 5
;l. a. r:ll
)
'@
*
; } :t
.
s
@
.j
.e
fi
@
S11
2
f;
,',."
Xlv j
jT
*
i
< N ;W:
UP
Figure 1. Reactivity with antibody to cytokeratins (Cam 5.2) (X250). A: Normal adrenal cortex, positive reactivity; B: Adrenal cortical adenoma (case 3). Note strongfocal reactivity; C: Adrenal cortical adenoma (case 1). Note weak staining offew cells. D: Adrenal cortical carcinoma (case 12). No reactivity seen. E: Renal cell carcinoma, positive reactivity. Figure 2. Reactivity with antibody to vimentin (X250). A: Normal adrenal cortex, staining confined to stromal elements; B: Adrenal cortical adenoma (case 3). Stainingprimarily confined to stromal elements: C: Adrenal cortical adenoma (case 1). Note strong reactivity; D: Adrenal cortical carcinoma (case 12), positive reactivity.
1082
Cote et al
AJPMay 1990, Vol. 136, No. 5
..
P
.
t4i; !
11~~~
-1
,
.! OF
-
,
k..~~U
&
-i-.t1f4
Figure 3.
Reactiv)ity
with antibodies to blood group
adrenal cortical carcinoma,
no
antigens (Lewis X) (X250). A: Adrenal cortical adenoma, reactivity seen; C: renal cell carcinoma, positive reactivity.
all expressed vimentin. There is a range of expression of both cytokeratins and vimentin in ACA, a characteristic not seen in normal cortex or ACC. In ACA, this range of intermediate filament expression appears to be coordinated in individual lesions. Adrenal cortical adenomas that show strong reactivity for cytokeratins only weakly express vimentin, while those that are weakly reactive for cytokeratins show strong expression of vimentin (Table 6 and Figures 1 B, 2B and 1C, 2C). These results suggest that adrenal cortical cells undergo characteristic changes in intermediate filament expression during the process of neoplastic conversion and malignant transformation. Our data would indicate that neoplastic conversion is associated with increased vimentin expression and decreased cytokeratin expression at the cellular level. Furthermore, it appears that using the techniques described here, lack of cytokeratin expression and strong vimentin expression is associated with malignant lesions. It is well known that malignant tumors of epithelial origin can coexpress both cytokeratins and vimentin.152 Furthermore, increased vimentin expression is associated Table 6. Comparison of Cytokeratin and Vimentin Expression in Adrenal Cortical Adenomas Cam 5.2 Case Vimentin* 1 + few cells +++ many cells 2 ++ focal +-++ focal +++ focal - most cells (+ stroma) 3 4 ++ focal + focal ++-+++ focal - most cells (+ stroma) 5 + focal ++ focal 6 Staining intensity: +, weak; ++, moderate; +++, strong; * stromal/ vascular elements stained in all cases.
no
reactivity
seen; B:
with more highly anaplastic epithelial tumors and anaplastic areas within these tumors, and thus it is correlated with certain morphologic features related to more aggressive behavior."5' However, we are not aware of any previous study or system in which the coordinated changes in intermediate filament expression described here were observed throughout the range of neoplastic growth. In addition, conversion from cytokeratin to vimentin expression is normally associated with morphologic changes in most tumors, primarily of the metaplastic or spindle cell type.15- " This was not the case for adrenal cortical neoplasms, where increased vimentin and decreased cytokeratin expression was not associated with any characteristic morphologic changes. The antigenic phenotype of adrenal cortical hyperplasia appears to be distinct and unrelated to this neoplastic conversion because none of four hyperplastic nodules studied had detectable expression of either cytokeratin or vimentin (unpublished data). As previously noted, cytokeratin expression is principally seen in the zona glomerulosa, while vimentin is not seen well in any layers of the adrenal cortex. This may indicate that adrenal cortical hyperplasia (those studied here at any rate) may arise from cells distinct from those that give rise to adrenal cortical neoplasms, and that hyperplasia may not be,part of the pathway of transformation in the adrenal cortex. The second major issue investigated here was whether antibody panels could distinguish adrenal cortical lesions from RCC. These tumors can be quite similar morphologically, and they are included in the differential diagnosis of metastatic lesions that have clear or granular cytoplasm. In addition, the primary site of a tumor involving both the adrenal gland and upper pole of the kidney
Adrenal and Renal Cortical Tumors
1083
AJP May 1990, Vol. 136, No. 5
can be difficult to determine, even after detailed microscopic examination. Our results suggest that the most useful antibodies are those with specificity to cytokeratins and blood group antigens (Table 5). Adrenal cortical carcinomas did not express detectable levels of cytokeratins. Cytokeratins were demonstrated in all primary clear cell and tubular RCCs examined. The majority of primary RCCs express blood group antigens, but all ACAs and ACCs were negative (Table 5). Reactivity with antibodies to vimentin and protein S-100 do not contribute to the differential diagnosis. Although Wick et al14 found that ACCs and RCCs did not express detectable levels of protein S-100, we found that many ACCs and the majority of RCCs showed positive reactivity to the antibody to protein S-100. Expression of protein S-100 by normal and malignant renal epithelium has been previously reported.21 22 Wick et al14 also demonstrated that the majority of RCCs expressed cytokeratins as well. In addition, they found that RCCs expressed epithelial membrane antigen (EMA), while ACAs and ACCs were negative. They reported that RCCs reacted with antibodies to the major blood group isoantigens (A, B, and H),14 but in our series none of the RCCs tested expressed detectable levels of these antigens. However, the majority of primary RCCs expressed Lewis-type blood group antigens, most commonly Lewis X, and occasionally Lewis a or b.23 One possible reason for this discrepancy is that the antibodies to the major blood group antigens used in the previous study may have had unsuspected cross reactivity with Lewistype antigens as well. It is important to emphasize the fact that the antigenic phenotype for RCCs reported here pertains to primary renal cortical tumors with clear cell or granular cell features. We have found that metastatic RCC less commonly express Lewis blood group antigens.23 In addition, some metastatic RCCs express low to undetectable levels of cytokeratins and high levels of vimentin (Cordon-Cardo et al, unpublished data), thus making their antigenic phenotype more similar to that of ACC. However, RCCs that lack expression of cytokeratins are generally poorly differentiated tumors, often with pronounced spindle cell features,'4 and thus can generally be separated from ACCs on morphologic grounds. Finally, RCCs and adrenal cortical lesions can be distinguished by a series of antigens that are expressed by cells of proximal tubular origin.24' These antigens are, however, only stable in frozen sections and were not used in the current study. This study points out once again the importance of using antibody panels in evaluating lesions of uncertain histogenesis or malignant potential. While none of the antigens studied here are specific for adrenal cortical or renal cells, lesions of similar histogenesis (ACAs vs. ACCs)
and lesions of different histogenesis (adrenal cortex vs. renal cortex) can be distinguished on the basis of their composite antigenic phenotypes. An important future step will be to determine if those features that appear to distinguish ACAs from ACCs (primarily cytokeratin and vimentin expression) correlate with more established criteria (morphology, weight, and DNA content analysis), particularly for borderline lesions. We expect that, given more sensitive detection systems and a greater number of tumors studied, cytokeratin expression will be shown in at least a proportion of ACCs. However, the finding of strong vimentin expression associated with weak cytokeratin expression should raise the possibility of malignant potential in an adrenal cortical tumor.
References 1. Heinbecker P, O'Neal LA, Ackerman LV: Functioning and nonfunctioning adrenal cortical tumors. Surg Gynecol Obstet 1957,105:21-33 2. Huvos AG, Hajdu SI, Brasfield RD, Foote FW: Adrenal cortical carcinoma: Clinicopathologic study of 34 cases. Cancer
1969.25:354-361 3. Kay S: Hyperplasia and neoplasia of the adrenal gland. Pathol Annu 1976,11:103-139 4. King DR, Lack EE: Adrenal cortical carcinoma: A clinical and pathologic study of 49 cases. Cancer 1979,44:239-244 5. Weiss LM: Comparative histologic study of 43 metastasizing and nonmetastasizing adrenal cortical tumors. Am J Surg Pathol 1984, 8:163-169 6. Schteingart DE, Oberman HA, Friedman BA, Conn JW: Adrenal cortical neoplasms producing Cushings syndrome. Cancer 1968, 22:1005-1013 7. Tang CK, Gray GF: Adrenal cortical neoplasms: Prognosis and morphology. Urology 1975, 5:691-695 8. Hough AJ, Holifield JW, Page DL, Hartmann WH: Prognostic factors in adrenal cortical tumors: A mathematical analysis of clinical and morphologic data. Am J Clin Pathol 1979, 72: 390-399 9. Amberson JB, Vaughn ED, Gray GF, Naus GJ: Flow cytometric analysis of nuclear DNA from adrenal cortical neoplasms: A retrospective study using paraffin-embedded tissue. Cancer 1987, 59:2091-2095 10. Taylor SR, Roederer M, Murphy RF: Flow cytometric DNA analysis of adrenocortical tumors in children. Cancer 1987, 57:2059-2063 11. Hsu S-M, Raine L, Fanger H: Use of avidin-biotin-peroxidase complex (ABC) in immunoperoxidase techniques: A comparison between ABC and unlabelled antibody (PAP) procedures. J Histochem Cytochem 1981,29:577-580 12. Cordon-Cardo C, Finstad CL, Bander NJ, Melamed MR: Immunoanatomic distribution of cytostructural and tissue-associated antigens in the human urinary tract. Am J Pathol
1987,126:269-284 13. Miettinen M, Lehto VP, Virtanen I: Immunofluorescence microscopic evaluation of the intermediate filament expression
1084
Cote et al
AJP May 1990, Vol. 136, No. 5
14.
15.
16.
17.
18.
19.
20.
21.
of the adrenal cortex and medulla and their tumors. Am J Pathol 1985,118:360-366 Wick MR, Cherwitz DL, McGlennen RC, Dehner LP: Adrenocortical carcinoma: An immunohistochemical comparison with renal cell carcinoma. Am J Pathol 1986; 122:343-352 Medeiros LJ, Michie SA, Johnson DA, Warnke RA, Weiss LM: An immunoperoxidase study of renal cell carcinomas: Correlation with nuclear grade, cell type, and histologic pattern. Human Pathol 1988,19:980-987 Oberman HA: Metaplastic carcinoma of the breast: A clinicopathologic study of 29 patients. Am J Surg Pathol 1987,11: 918-929 Meis JM, Ord6onez NG, Gallager HS: Sarcomatoid carcinoma of the breast: An immunohistochemical study of six cases. Virch Arch 1987, 410:415-421 Santensanio G, Pascal RR, Bisceglia M, Costantino AM, Bosman C: Metaplastic breast carcinoma with epithelial phenotype and pseudosarcomatous components. Arch Pathol Lab Med 1988,112:82-85 Hurlimanh J, Gardiol D, Scazziga B: Immunohistology of anaplastic thyroid carcinoma: A study of 43 cases. Histopathology 1987,11:567-580 LiVolsi VA, Brooks JJ, Arendash-Durand B: Anaplastic thyroid tumors: Immunohistology. Am J Clin Pathol 1987, 87: 434-442 Vanstapel M-J, Gatter KC, Woolf-Peeters C, Mason DY, Desmet VD: New sites of human S-100 immunoreactivity detected with monoclonal antibodies. Am J Clin Pathol 1986,
85:160-168
22. Kato K, Haimoto H, Ariyoshi Y, Horisawa M, Washida H, Kimura S: High levels of S-100 protein in tumor tissues and in serum of patients with renal cell carcinoma. Jpn J Cancer Res 1985, 76:856-862 23. Cordon-Cardo C, Reuter VE, Finstad CL, Sheinfeld J, Lloyd KO, Fair WR, Melamed MR: Blood group-related antigens in human kidney: Modulation of Lewis determinants in renal cell carcinoma. Cancer Res 1989,49:212-218 24. Cordon-Cardo C, Bander NH, Fradet Y, Finstad CL, Whitmore WF, Lloyd KO, Oettgen HF, Melamed MR, Old LJ: Immunoanatomic dissection of the human urinary tract by monoclonal antibodies. J Histochem Cytochem 1984, 32:
1035-1040 25. Bander NH, Cordon-Cardo C. Monoclonal antibodies to renal cell markers. In Sell S, Reisfeld RA, eds. Monoclonal Antibodies in Cancer. Clifton, NJ, Humana Press, 1985. pp. 325-338. 26. Finstad CL, Cordon-Cardo C, Bander NH, Whitmore WF, Melamed MR, Old LJ: Specificity analysis of mouse monoclonal antibodies defining cell surface antigens of human renal cancer. Proc Natl Acad Scie USA 1985, 82:2955-2959
Acknowledgments The authors thank Drs. Myron Melamed and Murray Brennan for their comments. Daisy Jimenez-Joseph and Mary Ann Gangi provided expert technical assistance, Kin Kong and Lisa Hollis prepared the photomicrographs, and Grace Bell and Alice Haynes prepared the manuscript.
